How a Dubai 3PL cut document processing from 4 days to 18 minutes — without firing anyone.
Meridian Cargo Solutions moves 8,400 shipments a month across the UAE, Saudi, and Oman. 142 employees, roughly $14M ARR, mostly mid-market e-commerce and FMCG. Customs clearance, last-mile, reverse logistics — the full chain.
Our protagonist is Karim Haddad, Head of Operations. What kept him up: every Sunday night he'd open WhatsApp to 30+ messages from clients asking where their proof-of-delivery was, why an invoice was wrong, or why customs got delayed because the commercial invoice didn't match the bill of lading.
Every shipment generated 6 to 9 documents: commercial invoice, packing list, bill of lading, certificate of origin, customs declaration, delivery note, POD. Each one passed through 7 hands on average — shipper rep → ops coordinator → customs clerk → finance → account manager → client → finance again for reconciliation.
The errors weren't exotic: typos in HS codes, mismatched weights between BoL and invoice, wrong currency conversions. Each one meant a customs hold, an angry client email, or a 3-week payment delay. Finance was carrying $1.8M in stuck receivables at any given time because of doc mismatches.
A single HS code typo on a 22-container shipment for Meridian's biggest FMCG client triggered a 9-day customs hold in Dammam. The client invoked the SLA penalty clause. Karim spent that weekend on calls instead of at his daughter's birthday.
Ripple effects: ops team on 11-hour days, two coordinators quit in Q1, finance close slipped from day 5 to day 12 every month.
Vynara built a document intelligence layer that sits between Meridian's email / WhatsApp intake and their TMS. Documents land, get classified, extracted, cross-validated against the shipment record, and either auto-posted or flagged with a specific reason.
Seven weeks from kickoff to production. Humans stayed in the loop for exactly two things — everything else moved without a coordinator touching it. Customs clerks shifted from data entry to handling the 8% of edge cases that actually needed a brain.
I stopped checking WhatsApp on Sundays. That's the metric nobody puts in a case study but that's the one that matters to me.
I was worried we'd have to let people go. We didn't. We doubled volume per person and finally took on the Saudi expansion we'd been pushing back for two years.
If your ops team is keying data from PDFs into another system, you have this problem — regardless of whether you call yourself logistics, manufacturing, or a law firm. Classify, extract, validate against a source of truth, route exceptions.